The Bedwell School lurks ominously on a valley's edge, notable if only because the 60's sex comedy "Here We Go Round The Mulberry Bush" filmed there. The boys and girls of Bedwell had a not unwarranted reputation for being "a bit hard"; standards were poor, and the school was generally acknowledged as performing pretty badly when it came to education, although most of the pupils had a jolly good time.
As time passed, Bedwell somehow transformed itself into a competent, technically oriented school, always improving in the ratings. It was amongst the first in the country to introduce SMP Maths, along with unusual subjects such as Electronics, and Computer Science. The new headmaster, a 7-foot man with a penchant for hauling small boys into the air by the lapels, played no small part in these changes.
Nearby Stevenage College of Further Education was the proud owner of a roomful of teletypes, several large machines which were used for chewing up little rectangles of cardboard, and a 10 CPS modem the size of a small suitcase. One of the teletypes was connected to the modem. The modem connected to the world, but mostly to the DECsystem-10 at Hatfield Polytechnic.
Hatfield's KA-10 served not only the Poly, but also the surrounding schools and colleges. This mighty room-sized machine boasted six tape drives, a huge 256K of core, over 30Mb of disk, and with this, ran about 90 time-shared processes during peak hours. It had a PDP-8 front end, which was capable of monitoring and overriding any of the 128 attached terminals, and the attention of a small group of hackers who found it an interesting place to hang out.
My introduction to computing was by way of a program written in BASIC, which allowed me to type in a short message and printed it out in large characters onto a teletype's paper-tape punch. This, unlike the rather tedious but infinitely more common programs to input several numbers and calculate their sum, grabbed my attention. With these humble but suspect beginnings, a year passed during which I mastered BASIC, the worst language ever invented, and BBC-X, a baroque simulated assembly language of no possible use. I subsequently familiarised myself with FORTRAN, Algol, COBOL, TECO and MIC, stubbornly read the TOPS-10 system call manual, and became addicted to MACRO-10 programming. I began writing serious software; my IRC-like chat program, email system, DECtape control daemon and exploitation of a disk quota bug gained me the not wholly welcome attention I evidently deserved. I met many fine people, and spent a lot of time online.
When the time came to leave Bedwell School, both Stevenage College and Hatfield Polytechnic seemed happy to have me continue my dabblings. Over many long nights I continued to develop my email, talk and bulletin board systems, and on the third rewrite decided to combine all three into the ambitious TAVERN (Tertiary Acme Verbalisation & Epistle Recall Network). With such an excellent acronym to work with, coding began in earnest; but soon, in what was to become a recurrent pattern in life, a job interrupted my work.
With glowing recommendations from various academics, I interviewed at, and was employed by, a hapless computer company called DIGICO. They built their minicomputers from board-level through microcode and assembler to application. The applications were indescribably dull. Why would anybody go to all this trouble to emulate a proprietary ICL protocol? Without an online editor or even a BASIC compiler? And as for the whippet-racing scoreboards, I simply couldn't bear to think about them. I hung out a little at the British Library in London, where a smart terminal with international character set support was being designed, and made a small effort to drag the company into the 70's, but I didn't last long at DIGICO.
Whilst at DIGICO and after, the DECsystem-10 still called me. Like a cat bringing birds, I cheerfully brought deficiencies, suggestions and bugs to the good people of Hatfield Polytechnic, with about the same mix of reactions. I applied for various jobs around London, mostly of a lowly Trainee Operator nature; people told me I was over-qualified, and some stayed in touch. Attending various deniable parties, I acquired the sources for various TOPS-10 system utilities, and began to experiment with these, often with dire consequences. I had reinvented myself as a bearded DEC-10 Systems Programmer, and was having a lot of fun. But alas, the urge towards fiscal remuneration was too strong to be resisted: I eventually succumbed.
With a whole DEC-10 to play with, I was really in my element. The homebrew graphics packages, feeble databases and overblown accounting systems were rather tedious, although their original sin was to be written in FORTRAN. Hacking at TOPS-10, installing patches, tweaking kernel internals: that was the real fun. It seemed I'd discovered my destiny, and soon progressed into collaborations with folk at other places, such as Edinburgh University and Rutherford Laboratories, the two other major SRC/ICF sites. The networking side of things began to heat up, as ANF-10 developed and early versions of ftp were conjured into existence.
But things were soon to change. Looking remarkably like a Borg starship painted bright orange, The Prime appeared on the horizon, and landed in the computer room. The acolytes trembled in fear, and many left to pursue lucrative careers. Soon it became apparent that this dreadful contraption, with it's attendant washing-machine disk drives, deprived of all but a small number of flashing lights and lacking the ability to play even simple tunes through the console, was intended to replace the majestic blue DECsystem-10.
No sane person would ever write an operating system in FORTRAN. No sensible computer hardware consists of a processor emulating another processor. No user interface uses printing keys instead of backspace and line erase.
However, The Prime had some vaguely interesting quirks. It came with sources, so soon I discovered the delights of PMA, The Prime's Gothic assembly language, and PL/P, a castrated version of PL/1 without dynamic allocation or even the slightest vestige of built-in I/O. I shortly began writing my own support functions, and came up with a collection of procedures which looked remarkably like the stdio library. I hadn't seen C, yet, mind. And since the useless junkpile also had nothing resembling a make utility, I ended up writing one of those too, all to assist me with the software I was really supposed to be producing: a spooled magnetic tape archival system, a disk-to-disk backup system, more accounting software, and several dozen other very boring miscellaneous programs.
Somewhere at the rear of The Orange Box, unbeknownst to me, strange happenings were, well, happening. A grey cable sprang from a small blue box in the corner of the room, went for a short detour beneath the floor tiles, and ended up preventing the closure of certain access panels. I was informed that this cable had some 9600 bauds contained within it, went all the way around the country, and spoke in a peculiar dialect called X.25. What I was witnessing was the birth of JANET, with which I was to strike up a fulfilling and long-lasting, if often traumatic, relationship.
JANET had some books. They were coloured, which was quite cheerful, although the same could not be said of the contents. While the rest of the world was busily dreaming up the Internet, I spent my time with these Coloured Books, which described the protocols used by JANET for terminal access, file transfer, and email, amongst other things. Under the influence of the Grey Book, basically a slight variation of RFC822, I found myself writing a mailer daemon. Using the facilities of the Blue Book, that which contained the Network Independent File Transfer Protocol, I designed and brought into operation an "automatic" updating system, which transferred gigabytes of software to the unfortunate sysadmins of the ever-increasing numbers of Primes connected to JANET.
I spent most of eight years with The Archiver, which ate some 20Gb of useless data at UMIST alone; six years, perhaps, with The Mail Server, which grew to encompass multi-homed aliasing, mailing lists, campus-wide routing, chronic overloading and all the paraphernalia associated with such things. I ended up looking after PRIMOS, all the networking software, most of the homebrew applications: happily babysitting a distributed environment which peaked with some 26 machines. Not content with this, I also began to learn C, so soon I was busily combining C, PL/1, FORTRAN and assembler into amazing Prime monstrosities.
Other small, dull boxes began to proliferate in Rotwang's Lab, and I struck up a casual acquaintance with DOS, which was (and remains) a joke. PERQs and Cifers, running an alien and unfathomable operating system called UNIX, showed some faint promise, but were simply too slow and primitive to be used.
For reasons best known to those with expertise in planning office space, UMIST decided to move their little Central Services group into the Computation Department. This rather signalled the end of an era.
The Science and Engineering Research Council Interactive Computing Facility (SERC/ICF) Prime Service (commonly known as the "SERC Primes") was gradually winding down, having contributed greatly to the cause of UK Academic Networking. X.25, The Books in Blue (JANET niftp), Green (JANET terminal access protocol), Yellow (ISO transport service), Grey (the mutant 822 protocol) and Red (the awful JTMP), X.28, X.29, The ISO 7-Layer Model, DNICs and Fast Select, PSS; what more could anyone ask for?
SERC decided that it would be nice to give us a Sun fileserver and a bunch of workstations, and this was installed alongside a pile of Vax hardware in Computation's machine room. Computation already had UNIX experience, and were interested in hooking their machines into JANET in some way.
After very little thought, the way turned out to be UUCP from The Vax into The Sun, which in turn talked in a serial port flavour of niftp to The Prime. This exercise was accompanied by a crash course in UNIX sysadmin, which much to my surprise I found quite easy, and UNIX culture, which I also found unexpectedly palatable. As anyone who knows their UNIX history probably won't tell you, UNIX was originally conceived because the Guru Founders weren't allowed (by Management) to buy a DECsystem-10; and both UNIX and PRIMOS are descendants of Multics. With the budding enlightenment of the history, something changed in my psychology, and in a matter of months all resistance evaporated. I became a convert to the One True Operating System.
An evaluation of various X servers and terminals was undertaken, and some work on IC design software under an ESPRIT contract came and went. The Sun sprouted connections; serial lines hummed with UUCP and asynchronous niftp, and an X.25 connection into JANET provided the link for UNIX Coloured Books to operate. An length of coax connected The Sun fileserver to the attendant workstations, and soon began to snake around the department. This was the Ethernet, over which strange foreign protocols were used, like TCP/IP, Yellow Pages and NFS. Although all this presented many amusing challenges, daily life consisted of sysadmin, user support, writing documentation, playing with Sunview, X-Windows and shell scripts, all the mysterious rituals normally undertaken by fledgling UNIX hackers. Usenet arrived in the form of a batched feed across the X.25 niftp link, and suddenly UMIST and The World were joined.
Although there was considerable resistance within the backwater enclave known as the UK, it gradually became apparent that the Internet was taking over the rest of the world. The local Ethernet spread across the UMIST campus, with The Sun providing heaps of protocol conversions between it and JANET's X.25-based networking. A small group of rebels began experimenting with routing IP packets across X.25, and the Shoestring Project was born. Slowly and reluctantly, Manchester appeared on the Internet, with DNS and telnet gradually being used alongside the awful NRS and a variety of PADs. The venerable niftp was supplanted by real ftp, and the ever-hungry NNTP began its usual rabid consumption of resources.
In the middle of all this, I did a fair impression of a whirling dervish after half a bottle of whisky and several snorts of speed. I was programming masses of C under both UNIX and DOS. I was still babysitting The Prime, which sadly continued to handle central email hub functions. I was both interested in networking and encouraged to hack at it by my users; already, on The Sun, I was switching mail for most of the UMIST campus, providing conversion mechanisms for most of the JANET protocols, and pushing the network as A Good Thing. But none of this was Real Work: what I was supposed to be doing was debugging data caches written in FORTRAN, writing C functions to emulate various Prime system calls, and composing documents about how to list directories.
As time passed I was involved in various different activities, providing UNIX and PC support, and learning all about Novell, which was infesting large numbers of hapless IBM-compatibles. I played the role of sysadmin, advisor, hardware roadie, committee sitter, information point lackey and frequent devil's advocate, got shifted around between several different offices, floors and buildings, and began to get a little fed up with it all.
It was fairly clear to me that the SERC weren't going to continue throwing scraps, and that my future lay in local infrastructure. Because of the relationship between our little group and UMIST generally, my ambitions were constantly frustrated due to lack of resources. I'd decided to move into networking, since it was pretty much what I was already doing. This took 3 years to achieve, at which time my SERC contract was on the point of termination, and I was on the verge of insanity. This probably explains why I chose to work at Manchester University's Computing Centre.
If a persistent user eventually does manage to enter the building in search of assistance, the interior is cunningly designed to make it unlikely that anybody with relevant information can be found, and also to make escape almost impossible. On the lower floor, signs containing the feared voodoo words "MCC" and "UMRCC" direct the hapless adventurer through a maze of twisty little passages, all alike, and reveal the torrid history of what was the University of Manchester Regional Computing Centre and subsequently became simply Manchester Computing Centre.
Deep within the bowels of this nightmare cavern lurks the Network Unit, a motley collection of unusual individuals and their associated cybernetic extensions.
The Network Unit is an Extended Internet Services Provider. It provides network infrastructure ranging from the humble dialup modem link to fibre-optic rings capable of spewing gigabytes of data per second around the Manchester area. On the University campus, it connects to and provides routing for hundreds of departmental networks, provides links to dozens of other networks throughout the North West of England, and looks after a lot of cables. As well as performing all the duties of a traditional ISP, the Network Unit engages in bleeding edge development and research, and is also notable for its effective handling of legacy systems.
Amongst the vast piles of mismatched equipment claimed by this unusual group, several unhappy and underpowered UNIX machines perform Tasks Vital to Mankind's Future, shovelling gigabytes of data around the network and conducting hundreds of simultaneous conversations with other machines throughout the world.
When I moved from UMIST to MCC, I brought not only my humble skills, but also the goodwill and cooperation of a fair number of loyal users. My particular UNIX experience was unique within MCC, which at that time was stumbling along the emulated lookalike route on its much-revered mainframe; nobody except myself had really used BSD UNIX, but the Network Unit, an island of sanity within the stormy SysV sea, was installing Suns. Whilst still overseeing the operation of both Sun and Prime at UMIST, I tightened up the existing Network Unit system and installed new hardware for my users and myself. At last I had a workstation on my desk!
Those services which were supplied by The Sun and The Prime at UMIST were largely superseded by similar services provided by my small collection of Suns. The central mail switching, Usenet, DNS services, X.25 gatewaying and a whole barrel-load of minor Internet services moved to the Network Unit, and from there, began to spread throughout the Manchester campus. A cycle of expansion and chronic overloading began, parallelling the growth of the Global Internet; more hardware and upgrades to existing systems were repeatedly absorbed by new services and a massive expansion in the user population.
At the time of writing, I'm mostly responsible for some dozen or so UNIX boxes, with lesser responsibilities and obligations to hundreds or perhaps thousands of other machines. I spend my time largely with SunOS, Solaris, Linux and FreeBSD, and dabble with Windows occasionally. Most of my work life is spent keeping services ticking over; despite redundancy and a certain degree of investment, the spectre of meltdown is never far away.
Just for the record, this is the most boring load of twaddle I've ever written, and I won't do it again.
Author: Ian Pallfreeman: ip@xenopsyche.com
Sometime in 1995.